You Ask, I Answer: Is AI Killing Art?

In today's episode, Anna asks, watching everyone make art and logos via AI, portraits from photos, trace via procreate, etc. Brings it home. Is art dead because people just upload whatever they want or ask a machine to do it. I mean, no. Having machines creating art is two things.

First, machines cannot create original art. Now, what they create, you know, tools like Wombo and Dolly and stuff like that, yes, they're creating things that um don't necessarily exist in that form before, but they are basing it on training data. Like to understand the answer to this question, you have to understand how AI works. AI works by taking large amounts of existing data and treat teaching machines to recognize it, classify it, categorize it, and then create new stuff from that. Right.

So if you give machines a bunch of ingredients, it can make new combinations of those ingredients. Sometimes in which something's been created that has never existed before. If the machine has never learned it from an existing piece of data, it can't create versions of that. Now, where these tools do interesting things is because they are trained on very large data sets that can be maybe the sum total of a type of knowledge or creations for a genre or niche or vertical. Yes, they can create things that are novel in that vertical, but they can't create anything that is brand new, net new, never seen before.

You'll see this problem even without AI in the movie industry. If you spend some time listening to soundtracks from movies, you will notice that there's not a ton in mainstream Hollywood that's new, right? And in fact, there's a great YouTube video series about this, but in fact, it has gotten so bad, the lack of originality that what's been created is unoriginal and not particularly noteworthy. Try this for an example. This is a great exercise.

Again, I remember seeing this on YouTube. I can't remember which channel it was. Can you think of the theme to the movie Superman? Any of them? Can you think of the theme to Star Trek or to Star Wars?

Right? Can you think of the theme to any Batman movie? Probably, right? You can probably think of some of those themes. Can you think of the theme to Jurassic Park?

Very memorable pieces of music, original pieces of music. Can you think of a theme to a uh Marvel uh uh Marvel Cinematic Universe movie? Like Captain America or The Avengers, uh, and like the 15 versions of that, or the Doctor Strange movie. Can you can you remember that music? I can't.

I've seen these movies a bunch of time. There's nothing memorable about that music. Why? Because in a lot of cases, it's derivative. It sounds okay in the moment, but it's not original.

And this is the issue that AI is running into. It can create stuff that's good. It can't create anything that's great. It can't create anything that is right now so novel and memorable and powerful that it embeds itself in our psyches. It will get there probably, but right now that's not the case.

So is art dead? Mediocre art is absolutely dead. Mediocre art. Mediocre art meaning it's it's it's not particularly good, it's not at all creative. Um, it's not at all original, it's purely derivative.

Yeah, that's dead because a machine can do that. People have been doing that. Is original art dead? Nope. Is creative art dead?

Nope. Because the machines are still struggling with this. And the nature of artificial intelligence makes this very difficult for machines to do, right? Machines inherently cannot train on data that does not exist. So something that is unique and new and creative can't be something that is in a training data set, because by definition, it's not new.

That's a ringing endorsement. Yeah, that was pretty okay. Um, but it's better than what some of us can do without machine help. And this is what the part of AI that makes me very optimistic is it augments our intelligence, it augments our capabilities. I am not a particularly good artist.

I am not at all a musician, right? Um, and yet I can give ideas and a rough framework or you know, a rough starting point to a machine, to uh to some AI, and it can create something that is better than what I would do. So my ambitions, my creative ambitions, maybe you make this magical, wonderful landscape painting, but all I can do is like make a mess in Microsoft Paint, right? My ambitions can be matched by a machine and the talent it brings, as long as I'm providing the framework for you know feedback to it to say like, yes, this is what I want to know, this is what I don't want. And that to me is very powerful because the more we enable people to express themselves, the the happier people will be.

If you've got this idea for the great American novel and you can't write, you're just not a good writer. Machines can help. We're not at the point yet where they can just write out you just you just give it a prompt and they write out something coherent that is good, that is aligned with what you want. That's gonna take some more time. But we're getting to the point where you can bang away a few sentences in a tool like one of the GPT models, and the machine can autocomplete a bunch of stuff afterwards and speed you on your way towards writing that that great novel that you wanted to write that you've that you think has been living inside you and has just been languishing because you're you're struggling to to put the words together.

Um machines can help you construct music. Is it exactly what it you know the an original piece composed by someone with true musical talent create? No, absolutely not. Is it better than what I would do face rolling on a piano keyboard? A hundred percent, right?

It when you listen to machine generated music that's been guided by a human, it's it's pretty okay. Right? It's it's in Marvel Cinematic Universe territory. It's not memorable. It's not original.

It's not particularly compelling, but it's better than nothing. It's better than, you know, face smashing into the piano. So art is not dead. AI is not killing art. Not true original art.

It is augmenting people's capabilities who don't have skills. And it is cranking out stuff that's mediocre. The question we have to ask ourselves is at what level of excellence are we do we find acceptable, right? Is a machine painted landscape acceptable? Depends, I guess, on on what your tastes are um if your tastes are very sophisticated, you would say, no, that's not good enough.

I need a piece of art created by a true human artist that can f understand the nuances of the big picture of the art I want to create. On the other hand, if you're looking for a background for a poster on a restroom stall, maybe you're like, yeah, machine generated is fine. It's good enough for this application. So a big part of this the answer to this question is depends on what your needs are and what your standards are and whether your standards are above or below what a machine can create. Now you can see the outline here of a of a bit of a painting that I did with a squirt gun, right?

I'll move the camera. That's not particularly compelling art. Right? It's it's it's okay. It's fine.

It's good enough for the wall in my basement. But a machine could do that pretty easily. And it would do it as well or better than me. And that's okay. It this meets the needs.

Would this be appropriate for a gallery? No. Was it appropriate to sell for $5,000? No. Or make an NFT of it, I guess.

No. That's not it it's not good enough for that. So that's the answer to the question. AI is not killing great art. AI is not killing original art, unique art.

AI is a hundred percent replacing mediocre or bad art. And it should because again any repetitors task like that is better off for a machine to do it anyway. So really good question. Thanks for asking if you like this video go ahead and hit that subscribe button.